DO FAITH-BASED PRISONS WORK?

Transcription

1 DO FAITH-BASED PRISONS WORK? Alexander Volokh * ABSTRACT This Article examines everything we know about the effectiveness of faith-based prisons, which is not very much. Most studies cannot be taken seriously because they are tainted by the self-selection problem. It is hard to determine the effect of faith-based prison programs because they are voluntary, and volunteers are more likely to be motivated to change and are therefore already less likely to commit infractions or be re-arrested. This problem is the same one that education researchers have struggled with in determining whether private schools are better than public schools. The only credible studies done so far compare participants with nonparticipants who volunteered for the program but were rejected. Some studies in this category find no effect, but some do find a modest effect. But even those that find an effect are subject to additional critiques: for instance, participants may have benefited from being exposed to treatment resources that non-participants were denied. Thus, based on current research, there is no strong reason to believe that faith-based prisons work. However, there is also no strong reason to believe that they do not work. I conclude with thoughts on how faith-based prison programs might be improved, and offer a strategy that would allow such experimentation to proceed consistent with the Constitution. ABSTRACT I. INTRODUCTION II. THE SELF-SELECTION PROBLEM A. Naked Self-Selection Johnson s Brazil Study O Connor et al. s Theology Study Kerley et al. s Religiosity Study The Florida DOC s Kairos Horizons Study * Assistant Professor, Emory Law School, avolokh@emory.edu. I am grateful to Steve Aos, Scott D. Camp, Aimee Chin, Hashem Dezhbakhsh, Griffin Sims Edwards, Esfandiar Maasoumi, Marna Miller, Thomas P. O Connor, Elena Pesavento, Joanna E. Saul, George B. Shepherd, Joanna M. Shepherd, Christopher R. Taber, and Eugene Volokh. I am also grateful to Margaret Ada Sapozhnikov, Matthew E. Uretsky, and the librarians at Emory Law School for their able research assistance.

2 44 Alabama Law Review [Vol.63:1:43 5. Denny s Kairos Horizon Study Education Studies B. Studies with Some Controls La Vigne et al. s Florida Study Rose s Kainos Community Study Young et al. s Prison Ministry Study O Connor et al. s Lieber Prison Study Wilson et al. s COSA Study Self-Selection in Prisons and Schools C. Matching on the Propensity Score O Connor et al. s New York Study Johnson et al. s New York Study Camp et al. s Life Connections Program Study Education Studies III. POTENTIALLY VALID STUDIES A. The Roads Not Taken Instrumental Variables Exogenous Policy Shocks B. Using Rejected Volunteers The Texas InnerChange Studies OPPAGA s FCBI Study Hall s Putnamville Study Hercik et al. s Kairos Horizon Study Wilson et al. s Detroit TOP Study O Connor et al. s Detroit TOP Study Education Studies IV. CONCLUSION I. INTRODUCTION There are five things one should know about faith-based prisons: (1) There are a lot of them out there. As of 2005, eighteen states and the federal government had some sort of residential faith-based program, 1 aimed at rehabilitating participating inmates by teaching them subjects like ethical decision-making, anger management, victim restitution, 2 and substance abuse 3 in conjunction with religious principles. 1. U.S. Dep t of Justice, National Institute of Corrections, Residential Faith-Based Programs in State Corrections, SPECIAL ISSUES IN CORRECCTIONS, Sept. 2005, pdf. 2. Programs, U.S. DEP T OF JUST. ARCHIVE, rams.html#3 (last visited Oct. 26, 2011). 3. Ams. United for Separation of Church & State v. Prison Fellowship Ministries, 509 F.3d 406, (8th Cir. 2007).

3 2011] Faith-Based Prisons 45 (2) One of them the InnerChange Freedom Initiative program in Iowa was struck down on Establishment Clause grounds in 2006, 4 but various faith-based prison programs still exist, 5 including InnerChange programs in other states. 6 InnerChange programs, which are explicitly motivated by Christian and Biblical principles, 7 are probably more vulnerable to constitutional challenges; programs that are more interfaith and have less explicitly religious content, like Florida s Faith- and Character-Based Institutions 8 or the federal Life Connections Program, 9 are probably less so. (3) Faith-based prisons continue to be promoted as promising avenues for reform, chiefly on the grounds that they improve prison discipline and reduce recidivism. 10 (4) However, most of the empirical studies of the effectiveness of faithbased prisons have serious methodological problems and, to the extent they find any positive effect of faith-based prisons, can t be taken at face value. (5) Those few empirical studies that approach methodological validity either fail to show that faith-based prisons reduce recidivism or provide weak evidence in their favor. In what follows, I explain and critically evaluate the empirical studies of the effectiveness of faith-based prisons. The reader who gets through this Article will know everything that we currently know about whether they work, by which I mean, chiefly, whether they reduce in-prison infractions or some measure of post-release recidivism, such as time to re-arrest, probability of re-arrest, or probability of reconviction. As the summary above indicates, we don t know much about the effectiveness of faith-based prisons. This is a shame, because the empirics of faith-based prisons are important, both to the legal community and to policymakers generally. 4. Ams. United for Separation of Church & State v. Prison Fellowship Ministries, 432 F. Supp. 2d 862, 933 (S.D. Iowa 2006), aff d, 509 F.3d 406, 425 (8th Cir. 2007). 5. E.g., Florida Expanding Faith-Based Prisons, UPI.COM (May 11, 2009, 9:41 AM), upi.com/top_news/2009/05/11/florida-expanding-faith-based-prisons/upi /. 6. E.g., Programs, INNERCHANGE FREEDOM INITIATIVE, (last visited Oct. 26, 2011). 7. Ams. United for Separation of Church & State, 509 F.3d at Faith- and Character-Based Institutions, FLA. DEP T OF CORRECTIONS, us/oth/faith/ci.html (last visited Oct. 11, 2011). 9. U.S. DEP T OF JUST. ARCHIVE, supra note See, e.g., Marti W. Harkness, Staff Dir., Fla. Legislature Office of Program Policy Analysis & Gov t Accountability, Review of the Dep t of Corrections Faith-Based Prisons (Jan. 25, 2011),

4 46 Alabama Law Review [Vol.63:1:43 First, we should care about the empirics because faith-based prisons and rehabilitative services are, and will continue to be, on the cutting edge of Establishment Clause litigation, 11 and empirics matter in the law. One might think that whether a program works shouldn t matter to whether it s consistent with the Establishment Clause cases; but in fact, there are several areas in Establishment Clause doctrine that seem to allow targeted uses of statistical evidence. 12 In any event, regardless of whether empirics should matter in Establishment Clause cases, reading judicial opinions suggests that they do. In Americans United for Separation of Church & State v. Prison Fellowship Ministries, which invalidated Iowa s contract with the InnerChange Freedom Initiative, District Judge Robert Pratt, immediately before launching into his Establishment Clause analysis, concluded his description of the faith-based program with the following complaint: More significant [than the warden s personal testimony about the program s beneficial in-prison effect], however, is the lack of evidence presented by the Defendants about the effect of InnerChange on recidivism. Aside from anecdotes, the Defendants offered no definitive study about the actual effects the InnerChange program has on recidivism rates. [The warden s] predecessor... communicated his desire early on in the initial RFP process that accountability for the program be included in the contractual agreement between the parties. Specifically, he requested at least annual program evaluations to include, but not limited to, re-incarceration rates and other measurable outcomes. But, in fact, there was no information presented at trial about whether InnerChange participants are more or less prone to recidivism than other inmates. 13 Of course, finding empirical language in opinions isn t definitive evidence that the empirics are relevant. Maybe judges who claim to care about the effectiveness of faith-based programs are just indulging in legally meaningless rhetoric designed to support a conclusion they already arrived at by strictly legal means. On the other hand, if judges discuss empirical data, 11. See, e.g., Ams. United for Separation of Church & State, 509 F.3d 406; Teen Ranch, Inc. v. Udow, 479 F.3d 403 (6th Cir. 2007); Freedom from Religion Found., Inc. v. McCallum, 324 F.3d 880 (7th Cir. 2003); Williams v. Lara, 52 S.W.3d 171 (Tex. 2001); see also Ira C. Lupu & Robert W. Tuttle, The Faith-Based Initiative and the Constitution, 55 DEPAUL L. REV. 1 (2005); Ira C. Lupu & Robert W. Tuttle, Zelman s Future: Vouchers, Sectarian Providers, and the Next Round of Constitutional Battles, 78 NOTRE DAME L. REV. 917 (2003); Ira C. Lupu & Robert Tuttle, Sites of Redemption: A Wide-Angle Look at Government Vouchers and Sectarian Service Providers, 18 J.L. & POL. 539 (2002). 12. In a voucher context, they could help establish that a choice was genuine. In a coercion context, they could help establish whether there was subtle pressure to join a program, or whether instead participants joined a program because of valid quality reasons. Effectiveness studies could help establish a secular purpose in cases where the purpose is doubtful. Cf. Perry v. Schwarzenegger, 704 F. Supp. 2d 921, (N.D. Cal. 2010) F. Supp. 2d 862, 914 (S.D. Iowa 2006) (citation omitted).

5 2011] Faith-Based Prisons 47 chances are that they believe the extra rhetorical force is useful in persuading someone and thus makes their opinion more influential and less subject to reversal. 14 Moreover, if we put our Legal Realist hats on, 15 maybe it s the judges views of effectiveness that are driving their legal conclusions (at least for some judges, who aren t in favor of or against faith-based prisons on ideological grounds). At the very least, if judges believe that a program is effective, they may think it s a shame if the program were to be found unconstitutional and might therefore be extra careful in their legal analysis to avoid striking it down. (If we disagree with their conclusion, we might replace the word careful in the previous sentence with the word imaginative.) Conversely, if they believe a program is ineffective, they may feel no particular pressure to uphold it. Perhaps a better piece of evidence that the empirics matter is that lawyers spend time discussing empirical studies in their briefs. Perhaps in response to Judge Pratt s concern, when the case came up to the Eighth Circuit, the Alliance Defense Fund and others submitted an amicus brief arguing that InnerChange s faith-based rehabilitative prison programs are proven to reduce recidivism. 16 (Unfortunately, of the two studies cited favorably in their brief, one has serious methodological problems, 17 while the other, properly interpreted, shows no beneficial effect of the program. 18 ) Empirical data also seems important in other Establishment Clause contexts. In Zelman v. Simmons-Harris, which upheld an Ohio program that allowed publicly funded vouchers to be redeemed at religious schools, Justice Souter s dissent cited statistics on the academic performance of schools to judge whether the parents choices were genuine. 19 Justice O Connor s concurrence similarly used performance statistics to rebut Justice Souter s argument. 20 Justice Thomas, for his part, noted that the success of religious and private schools is in the end beside the point, because the State has a constitutional right to experiment with a variety of different programs to promote educational opportunity. 21 But that didn t stop him from marshalling statistics to argue that [r]eligious schools, like other private schools, achieve far 14. See, e.g., Alexander Volokh, Choosing Interpretive Methods: A Positive Theory of Judges and Everyone Else, 83 N.Y.U. L. REV. 769, (2008) (discussing how judges use rhetoric to, among other things, maximize persuasiveness and minimize reversal). 15. E.g., Oliver Wendell Holmes, The Path of the Law, 10 HARV. L. REV. 457 (1897). 16. Brief for Alliance Defense Fund et al. as Amici Curiae Supporting Defendant-Appellants at 15, Ams. United for Separation of Church & State, 509 F.3d 406 (No ), 2006 WL (citing JOHNSON & LARSON, infra note 282; Johnson, infra note 201). The Alliance Defense Fund was joined on this brief by the National Association of Evangelicals, the Center for Neighborhood Enterprise, Teen Challenge, Time to Fly, the Center for Public Justice, Evangelicals for Social Action, and the Coalition to Preserve Religious Freedom. 17. See infra text accompanying notes (discussing Johnson, infra note 201). 18. See infra text accompanying notes (discussing JOHNSON & LARSON, infra note 282) U.S. 639, n.10 (2002) (Souter, J., dissenting). 20. Id. at (O Connor, J., concurring). 21. Id. at 681 (Thomas, J., concurring).

6 48 Alabama Law Review [Vol.63:1:43 better educational results than their public counterparts. 22 That Ohio s program includes successful schools, Justice Thomas wrote, simply indicates that such reform can in fact provide improved education to underprivileged urban children. 23 Here, too, seasoned appellate litigators spent several pages of their Supreme Court brief discussing the history and performance of voucher programs 24 and this was in a party s argument, not just an amicus brief 25 though this discussion was ostensibly not to convince the Court that parental choice is proper public policy. 26 But perhaps more importantly, we should care about the empirics because, whether or not they should matter in the law, 27 they obviously should matter in policy. If faith-based prisons don t reduce recidivism, the case for their funding is correspondingly weakened (though they may still be supported by other arguments). 28 If they do reduce recidivism, or if they have other policy advantages, then even if judges are wrong to stretch the law to find them constitutional, we aren t wrong to try to find ways to allow them to function constitutionally. 29 After all, even the non-religious have an interest in the rehabilitation of prisoners, and if religion can play a positive role in rehabilitation, this may be good news not only to the irreligious but even to those who are hostile to religion. It is thus unfortunate that the legal literature hasn t done a good job evaluating the empirical evidence on faith-based prisons. The law reviews are devoid of any comprehensive, critical discussion of the existing studies. Most legal articles on the subject simply choose not to bother with empirical 22. Id. 23. Id. 24. Brief for Petitioners, Zelman, 536 U.S. 639 (No ), 2001 WL at * The seasoned appellate litigators were former Solicitor General and Harvard Law School professor Charles Fried, Institute for Justice litigators Clint Bolick and William H. Mellor, and Cleveland appellate litigator David Tryon. Id. 25. The other side of the empirical voucher debate in Zelman was, however, represented by amici. E.g., Brief for Nat l Sch. Bds. Ass n et al. as Amici Curiae Supporting Respondents at *9 14, Zelman, 536 U.S. 639 (Nos , , ), 2001 WL (relevant pages missing in Westlaw), also available at HarrisvZelmanUSSupCtdup.aspx. 26. Brief for Petitioners, supra note 24, at *25; cf. Brief of Respondents at 40, Zelman, 536 U.S. 639 (Nos , , ), 2001 WL , available at reme_court/briefs/ / mer.simmons.pdf. 27. Cf., e.g., Brief for Respondents, Zelman, supra note 26, at See infra note I discuss such a way briefly in the Conclusion. See generally Alexander Volokh, The Constitutional Possibilities of Prison Vouchers, 72 OHIO ST. L.J. (forthcoming December 2011). 30. See Lynn S. Branham, The Devil Is in the Details : A Continued Dissection of the Constitutionality of Faith-Based Prison Units, 6 AVE MARIA L. REV. 409, (2008); James A. Davids, Putting Faith in Prison Programs, and Its Constitutionality Under Thomas Jefferson s Faith-Based Initiative, 6 AVE MARIA L. REV. 341, (2008); Marc O. DeGirolami, The New Religious Prisons and Their Retributivist Commitments, 59 ARK. L. REV. 1, (2006); Alex J. Luchenitser, InnerChange : Conversion as the Price of Freedom and Comfort A Cautionary Tale About the Pitfalls of Faith-Based Prison Units, 6 AVE MARIA L. REV. 445, (2008); Nathaniel Odle, Privilege Through Prayer: Examining Bible-Based Prison Rehabilitation Programs Under the Establishment Clause, 12 TEX. J. C.L. & C.R. 277, (2007).

7 2011] Faith-Based Prisons 49 studies. Some legal articles do address the empirical evidence, 30 but their discussions are generally quite cursory. 31 Outside of the legal literature, a few review articles do take a broader approach. 32 Some even single out which articles they believe are methodologically more valid than others. 33 But, unfortunately for other scholars, these generally have no in-depth discussion of the studies being reviewed, nor do they discuss why the studies identified as superior really are superior. The rest of us are thus left to either take them at their word (I myself disagree with some of their assessments) 34 or track down the studies (many of which are hard to find) and read them ourselves a daunting task for those without empirical training. 35 This Article fills the void. More specifically, this Article makes three distinct contributions. First, I provide a detailed critical discussion of (to my knowledge) every existing empirical study on the effectiveness of faith-based prison programs. The word every in the last sentence masks a few critical decisions. In making such a survey, I have chosen to include certain kinds of studies and exclude others. Because the relevant legal issues center around immersion -style faith-based prison units that seek to immerse prisoners in an almost monastic or total experience of religiously based living, 36 I exclude studies that explore more general issues like the effect of religiosity. 37 The 31. The most comprehensive law review treatment of the empirical studies discusses five different empirical studies of faith-based prison programs. Davids, supra note 30, at (citing Johnson et al., infra note 193; Johnson, infra note 201; JOHNSON & LARSON, infra note 282; JEANETTE HERCIK ET AL., DEVELOPMENT OF A GUIDE TO RESOURCES ON FAITH-BASED ORGANIZATIONS IN CRIMINAL JUSTICE 161 (2004), (discussing the results of HERCIK ET AL., infra note 307); id. at 32 (citing O Connor et al., infra note 342)). But it accepts their positive conclusions largely uncritically, whereas only three of these studies approach methodological soundness. (I discuss JOHNSON & LARSON, infra note 282, HERCIK ET AL., infra note 307, and O CONNOR ET AL., infra note 331, in Part III infra.) Moreover, of those three, two, properly interpreted, show no effect, see infra text accompanying notes 282 and 307, and the third shows, at most, weak effects. See infra text accompanying note 331; see also infra Part IV (resources problem). 32. E.g., Daniel P. Mears et al., Faith-Based Efforts to Improve Prisoner Reentry: Assessing the Logic and Evidence, 34 J. CRIM. JUST. 351 (2006). 33. E.g., STEVE AOS ET AL., EVIDENCE-BASED ADULT CORRECTIONS PROGRAMS: WHAT WORKS AND WHAT DOES NOT (Wash. State Inst. for Pub. Policy, Jan. 2006), pdf; Thomas P. O Connor, What Works, Religion as a Correctional Intervention: Part II, 14 J. CMTY. CORRS. 4, Winter , available at whatworks2.pdf?ga=t. 34. Tom O Connor, see O Connor, supra note 33, is himself a contributor to the literature and evaluates the methodology of some of his own studies. I disagree with O Connor s assessment of one of the studies, which he ranks good, see O Connor et al. infra note 65, but which I lump in with the studies marred by naked self-selection. I also disagree with some of Aos s assessments. See AOS infra notes 111, 135, & Several of these studies are hard to find, and I ve had to obtain them directly from the authors. 36. Thomas P. O Connor & Jeff B. Duncan, Religion and Prison Programming: The Role, Impact, and Future Direction of Faith in Correctional Systems, 11 OFFENDER PROGRAMS REP. 81, 86 (2008); see also Thomas P. O Connor et al., Criminology and Religion: The Shape of an Authentic Dialogue, 5 CRIMINOLOGY & PUB. POL Y 559, 563 (2006) (similar). 37. Todd R. Clear et al., Does Involvement in Religion Help Prisoners Adjust to Prison?, NCCD FOCUS, Nov (measuring religiousness by the Prisoner Values Survey); Todd R. Clear & Marina Myhre, A Study of Religion in Prison, 6 INT L ASS N RES. & CMTY. ALTS. J. ON CMTY. CORRS., no. 6, 1995 at 20 (attitudinal measures of religion); Todd R. Clear & Melvinia T. Sumter, Prisoners, Prison,

8 50 Alabama Law Review [Vol.63:1:43 measures of religiosity, in these studies, are either general measures of how religious an inmate feels or how many times an inmate attends religious services in prison. These studies may be valuable for some purposes, but they don t help in evaluating immersion-style faith-based units since ordinary prison worship services are both widespread and uncontroversial. I instead focus on studies of the effectiveness of specific faith-based interventions on variables of interest like the likelihood of recidivism. I also include studies of religious after-care for released inmates, even though these aren t technically in-prison programs. 38 Second, I provide a detailed discussion of the methodological issues involved in evaluating faith-based prisons generally. In statistics, methodology is everything; it s a shame that the legal community, which often relies on these empirical studies, isn t as sophisticated as it could be at telling valid studies apart from invalid ones. Roughly speaking, the studies fall into four categories. Three of them naïve comparisons of participants to non-participants, comparisons with some controls, and matching based on propensity scores aren t credible because they don t account for what is, in my view, the most serious obstacle to effective assessment: the self-selection problem. Inmates who are motivated enough to choose to participate in a rehabilitative program are already less likely to reoffend. So any study that compares voluntary participants and voluntary non-participants may just be picking up the effect of being a good person, not the effect of the program itself. (Some of these and Religion: Religion and Adjustment to Prison, 35 J. OFFENDER REHAB., nos. 3 & 4, 2002, at 125 (measuring religiousness by the Prisoner Values Survey); Byron R. Johnson, Religiosity and Institutional Deviance: The Impact of Religious Variables upon Inmate Adjustment, 12 CRIM. JUST. REV. 21 (1987) (using self-reported religiosity, chaplain-reported religiosity, and church attendance); Byron R. Johnson et al., A Systematic Review of the Religiosity and Delinquency Literature: A Research Note, 16 J. CONTEMP. CRIM. JUST. 32 (2000) (meta-analysis concentrating on church attendance, prayer, religious salience, and Bible study, though including some other religious activities ); Thomas P. O Connor & Michael Perreyclear, Prison Religion in Action and Its Influence on Offender Rehabilitation, 35 J. OFFENDER REHAB., nos. 3 & 4, 2002, at ( rate of religious participation doesn t distinguish between ordinary services and retreats); Michael G. Pass, Religious Orientation and Self-Reported Rule Violations in a Maximum Security Prison, 28 J. OFFENDER REHAB., nos. 3 & 4, 1999, at 119 (attitudinal measures of religion); Thomas P. O Connor, A Sociological and Hermeneutical Study of the Influence of Religion on the Rehabilitation of Inmates (2003) (unpublished Ph.D. dissertation, Catholic University of America) (on file with author) (count of religious attendance generally); Melvina T. Sumter, Religiousness and Post-Release Community Adjustment (1999) (Ph.D. dissertation, Florida State University), (using Prisoner Values Survey and similar attitudinal variables). 38. I have also excluded programs whose religiosity is unclear. Transcendental Meditation, for instance, is considered religious by some and non-religious by others. Compare, e.g., Amy Karasz & Meaghan Midgett, Transcendental Meditation, movements.lib.virginia.edu/nrms/tm.html (last modified Jan. 12, 2001), with The Technique, THE TRANSCENDENTAL MEDITATION PROGRAM, (last visited Oct. 26, 2011). Cf. Debbie Elliott, At End-of-the-Line Prison, An Unlikely Escape, NPR, /02/08/ /at-end-of-the-line-prison-an-unlikely-escape (last visited Oct. 26, 2011) ( The Vipassana technique, though secular, is based on the teachings of Buddha. ). In any event, Transcendental Meditation is quite different from the other sorts of religious programs canvassed in this survey. Nonetheless, a number of studies purport to find that Trancendental Meditation reduces recidivism. Below, see infra note 161, I cite two studies that belong in the Studies with Some Controls Part that is, studies that do not adequately control for self-selection.

9 2011] Faith-Based Prisons 51 studies are subject to even further sources of bias. For instance, in addition to self-selection in the decision whether and how intensively to participate, there can be selection by the program staff in the decision of whom to admit or whom to kick out, 39 as well as success bias in the consideration only of those who completed the program without dropping out. 40 ) In my view, the only credible studies so far fall into a fourth category those that compare (voluntary) participants in faith-based programs with people who volunteered for the program but were rejected. 41 And even these studies are subject to the resources problem : they compare participation in the program either with the alternative of no program at all or with the business as usual alternative of whatever other programs happened to be available, rather than with participation in a comparably funded secular program. Thus, even if a religious program is better than nothing at all, it could be because of the greater access to treatment resources for instance, mentors and counselors and not because of the religious content of the program. 42 Third, I set the empirical debate on faith-based prisons side by side with a parallel empirical debate: whether private schools are better than public schools. One striking aspect of the faith-based prisons research is how much it looks like the private school research. (Some of this research studies not private schools as such, but Catholic schools, since most private schools are religious, and Catholic schools are a homogeneous enough group to be susceptible to generalization.) Both faith-based prisons and private schools are subject to self-selection any naïve comparison between, on the one hand, secular prisons or public schools and, on the other hand, faith-based prisons or private/catholic schools is subject to the critique that the private or religious options have to be affirmatively chosen, and people who are motivated enough or whose parents are motivated enough to make that choice are already more likely to be high-achieving students or lowrecidivism inmates. 43 While sophisticated researchers in both areas are aware of the selfselection problem, the education literature has addressed it far more con- 39. E.g., the Brazil study, infra text accompanying notes 52 64; the theology study, infra text accompanying notes 65 68; and the Texas InnerChange studies, infra text accompanying notes E.g., the Brazil study, infra text accompanying notes 52 64; the Florida DOC s Kairos Horizons study, infra text accompanying notes 77 84; the Texas InnerChange studies, infra text accompanying notes ; Wilson et al. s Detroit TOP study, infra text accompanying notes ; and O Connor et al. s TOP study, infra text accompanying notes There are other promising statistical techniques, but so far they apparently haven t been used in the faith-based prison context. See infra Part III.A. 42. See infra Part IV. 43. An analogous problem occurs in various studies of the effectiveness of private prisons. A comparison of public to private prisons is biased if the private prisons studied, rather than getting a random sample of inmates, were sent a sample of inmates that was systematically healtheir or less dangerous than average. See, e.g., Richard A. Oppel Jr., Private Prisons Found to Offer Little in Savings, N.Y. TIMES, May 18, 2011 ( It s cherry picking, said State Representative Chad Campbell, leader of the House Democrats. They leave the most expensive prisoners with taxpayers and take the easy prisoners. ).

10 52 Alabama Law Review [Vol.63:1:43 scientiously and has progressed further than the prisons literature in advancing our empirical knowledge. Perhaps prison researchers could learn something from school researchers example. In the end, this Article has bad news and good news. The bad news, as I ve said above, is that most studies are low quality, and the results of the higher quality studies aren t promising. There seems to be little empirical reason to believe that faith-based prisons work. The good news is that there s also no proof that they don t work. The absence of statistically valid or statistically significant findings isn t the same as the presence of negative findings. And while the self-selection problem is real and important, the resources problem may not even be a problem at all: maybe the zero alternative or the business as usual alternatives really are proper empirical baselines, since they reflect both reality and, perhaps, political feasibility. So the picture isn t uniformly bleak: there are some programs that seem to show some statistically significant effects, even if they are weak and even if we re not sure how well they compare to the hypothetical effects of a hypothetical, comparably funded secular program. Perhaps future research will shed light on these questions. In the meantime, clearly some groups want to have such prisons, some inmates want to attend them, and they probably do little if any harm. If some programs don t work, this is an indication to future practitioners that something needs to be changed; if some programs work, maybe they can be replicated elsewhere. Better results won t emerge unless they re allowed to emerge by a process of experimentation. As I ve suggested above, it would be a shame if this process is cut off for constitutional reasons, provided there is a constitutionally valid way for them to proceed. At the end of this Article, I suggest such a way. Faith-based prisons, as currently constituted, are probably unconstitutional under modern Establishment Clause doctrine. But they would become fully constitutional under a system of prison vouchers that would allow inmates to choose their own prisons, whether secular or religious. I develop this idea at length elsewhere, 44 but the bottom line is that, despite the current weak evidence, some version of faith-based prisons may still work, and there is a way for that version to emerge consistent with the Constitution. II. THE SELF-SELECTION PROBLEM The most serious problem with studies of the effectiveness of faithbased prisons is the self-selection problem. Prisoners obviously select into faith-based prisons voluntarily. 45 And the factors that would make an inmate 44. Volokh, supra note At least in the United States, this would be so clearly required by the Establishment Clause that, even if the programs are ultimately found to be unconstitutional, this is a feature that the designers of such programs would be sure to include.

11 2011] Faith-Based Prisons 53 select a faith-based prison may also make him less likely to commit crimes in the future. 46 One such factor might be religiosity itself. 47 In addition, an inmate who takes the trouble to choose to join a rehabilitative program may be more motivated and more open to change, and this may itself make him more likely to change regardless of whether the program actually works. The following Parts illustrate three types of studies that don t adequately control for self-selection, both for faith-based prisons and for the analogous context of private/catholic schools. 48 The first type of study shows the self-selection problem in its most naked form: it simply compares the results of participants in a faith-based program with those of non-participants. The second type of study accounts for some of the differences between participants and non-participants by comparing the group of participants with a matched group of non-participants, where the matching is based on various observable factors like race, age, criminal history, and the like. 49 But, of course, such a procedure can t control for unobservable variables, like motivation to change. The third type of study uses a more sophisticated statistical technique called propensity score matching. Participants are matched to nonparticipants not based on observable factors directly, but based on their propensity score, that is, their estimated probability of participating in the program. While propensity scores are a useful technique in some applications, they don t alleviate the self-selection problem in the faith-based prison context. A. Naked Self-Selection The studies in this section purport to find a positive effect of faith-based prisons based on comparing, say, recidivism rates of participants in faithbased units and prisoners in the general population or in different prisons. But these sorts of studies aren t credible because they make no effort to control for self-selection. Without knowledge of the selection process, there is no way to determine whether observed differences between program participants and comparisons are due to actual program effects or are an 46. Scott D. Camp et al., An Exploration into Participation in a Faith-Based Prison Program, 5 CRIMINOLOGY & PUB. POL Y 529, 534 (2006). 47. E.g., sources cited supra note 37. I don t want to oversell this point, since it takes no imagination at all to imagine religiously inspired violence; moreover, one observes associations between white prison gangs and forms of Christianity or neopaganism (Odinism or Asatru), and between black prison gangs and forms of Islam. 48. See supra text accompanying note Matching a method from epidemiology is commonly used in faith-based prison studies. In education studies, instead of matching, generally we have a large number of observations of participants and non-participants; no one is matched to anyone else, but the researchers control for observable factors in a statistical sense, by including those factors as independent variables in a regression.

12 54 Alabama Law Review [Vol.63:1:43 artifact of preexisting differences between the groups. 50 Rather than giving us the effect of faith-based prisons, these studies may be giving us the effect of faith-based prisoners Johnson s Brazil Study Byron Johnson compared recidivism among inmates in two Brazilian prisons: Humaitá, a faith-based facility, and Bragança, a secular facility with vocational training programs. 52 Data wasn t available for 46% of the inmates, though the data loss didn t differ significantly between the two prisons. 53 High-risk Humaitá inmates had significantly 54 lower recidivism 12% of the high-risk Humaitá inmates were re-arrested after three years, versus 38% of the high-risk Bragança inmates. 55 The average number of rearrests was also significantly lower for Humaitá prisoners 56 even though on average the original offenses of the Humaitá prisoners had been more serious, they were more likely to be violent, and they had possibly served more time in prison. 57 The main problem with this study is that prisoners apply to be in Humaitá, prisoners families must be involved in the prisoner s recuperation process, prisoners aren t accepted without sufficient motivation and commitment to change, and prisoners don t stay unless they and the prison agree after an initial 60-day assessment period. 58 The results are thus tainted by multiple sources of bias: self-selection, selection by the prison itself, and success through the assessment period. Moreover, among low-risk inmates, recidivism rates weren t significantly different between the two prisons. 59 There was no significant differ- 50. Camp et al., supra note 46, at 529; see also James J. Heckman & Richard Robb, Alternative Methods for Solving the Problem of Selection Bias in Evaluating the Impact of Treatments on Outcomes, in DRAWING INFERENCES FROM SELF-SELECTED SAMPLES 63, (Howard Wainer ed., 1986) (explaining why a mere comparison of sample means does not yield the treatment effect when the decision to get treatment is correlated with unobservable variables). 51. Cf. Volokh, supra note 14, at 775 (because of failure to account for self-selection, many statements about textualism may really only be statements about textualists ). 52. Byron R. Johnson, Assessing the Impact of Religious Programs and Prison Industry on Recidivism: An Exploratory Study, TEX. J. CORR., Feb. 2002, at 7. O Connor rates this study as having fair methodological quality (on a poor-fair-good-excellent scale). O Connor, supra note 33, at 23 tbl.3. The study doesn t say how inmates arrived in the faith-based facility, but one suspects that they chose to be there. 53. Johnson, supra note 52, at In this Article, I use the term significant to mean statistically significant at the 5% level. 55. Johnson, supra note 52, at 9 tbl.1. The p-value of the differences between high-risk inmates (that is, the level of statistical significance) is less than 0.01 (that is, it s highly significant). Id. 56. Id. at 9. Humaitá prisoners averaged 0.23 re-arrests, compared to 0.53 for Bragança prisoners. Id. The p-value is less than Id. 57. Id. at ANGUS CREIGHTON, HUMAITA PRISON 5 (1998), brazil/humaita2/at_download/file; Jonathan Burnside, The Prison That Started It All, in JONATHAN BURNSIDE ET AL., MY BROTHER S KEEPER: FAITH-BASED UNITS IN PRISONS 1, 13 (2005). 59. Johnson, supra note 52, at 9 tbl.1. The p-value of the difference between re-arrest rates for lowrisk inmates is Id.

13 2011] Faith-Based Prisons 55 ence between times to re-arrest 60 or the severity of the subsequent offense. 61 The reincarceration rate was lower among Humaitá inmates, but the validity of this finding is questionable due to extensive data loss. 62 Moreover, many relevant background factors, like age or criminal history, weren t considered, perhaps because the data wasn t available. 63 Finally, Humaitá differs from other Brazilian prisons (possibly including Bragança) in many ways unrelated to religion. The environment is more pleasant, prisoners and their families are treated better, there are more (nonreligious) activities, and so on. 64 Any improvements in recidivism could therefore have been caused not only by selection, but also by better secular prison conditions. 2. O Connor et al. s Theology Study Thomas O Connor and his coauthors compared recidivism between 54 inmates who participated in a master s program in theology at Sing Sing prison and 402 non-participants. 65 Completion of the ministry program was associated with a significantly lower risk of re-arrest in the first 28 months out of prison only 9% of participants were re-arrested, compared to 37% of non-participants. 66 However, both self-selection and selection by program administrators taint these results. The students were selected by a highly competitive application and reference process ; the program was open only to inmates with a college degree, who read and wrote well, and who had references from chaplains and other inmates attesting to their religious commitment and showed a deep willingness to turn their lives around. 67 In fact, according to the president of the seminary that ran the theology program, the program had built-in success because they made sure to accept applicants 60. Id. at 9. Humaitá reoffenders were arrested, on average, 20.2 months after release, and Bragança reoffenders were arrested, on average, 18.3 months. Id. The p-value of this difference is Id. 61. Id. The severity ranking of the subsequent offense was 3.8 for Humaitá prisoners, vs. 3.0 for Bragança prisoners; the p-value of this difference was Id. 62. Id. Six of 10 Humaita inmates returned to prison, while 16 of 17 former prisoners from Braganca returned to prison. Id. This difference was significant at the 5% level based on a one-tailed Fisher s Exact Test. Id. 63. Gerry Rose, The Impact of Kainos and Christian-Based Units on Recidivism, in BURNSIDE ET AL., supra note 58, at 294, Burnside, supra note 58, at 14 tbl Thomas O Connor et al., Theology and Community Corrections in a Prison Setting, CMTY. CORR. REP., July-Aug. 1997, at 67, 68. O Connor (one of the authors of this study), in a separate work, rated this study as having good methodological quality (on a poor-fair-good-excellent scale). O Connor, supra note 33, at 23 tbl.3; see also TOM O CONNOR ET AL., RECIDIVISM AND THE MASTER IN PROFESSIONAL STUDIES PROGRAM AT SING SING PRISON: AN EXPLORATORY STUDY (Ctr. for Soc. Research, 1996). 66. O Connor et al., supra note 65, at 75. In a logistic regression analysis of re-arrest within 28 months, controlling for number of months out of prison, participation in the program had a coefficient of with a standard error of , which made it very highly significant, with a reported significance level of O CONNOR ET AL., supra note 65, at 7 tbl O Connor et al., supra note 65, at 67.

14 56 Alabama Law Review [Vol.63:1:43 who want to learn who they are, what they value and what they believe in Kerley et al. s Religiosity Study Kent Kerley and his coauthors examined the relationship between religiosity and negative prison behaviors at the Mississippi State Penitentiary in Parchman, Mississippi. 69 First, they measured inmates religiosity using a survey. 70 Most of these measures are irrelevant for our purposes because they don t involve specific programming 71 for instance, inmates were asked whether they had experienced a conversion and whether they believed in God. 72 But one of the measures was attendance at a one-day Prison Fellowship Ministries event called Operation Starting Line, which included Christian musicians, comedians, professional athletes, and other speakers, and which was held about six months before the survey. 73 Participation in Operation Starting Line predicted a significantly reduced rate of arguing with other inmates 52.5% of participants argued with other inmates once or more per month, as opposed to 60.0% of nonparticipants. 74 But participants and non-participants didn t differ statistically significantly in their likelihood of fighting once or more per month 18.9% for participants versus 19.3% for non-participants. 75 Inmates, of course, self-selected into the Starting Line events. In addition, the data was collected by a survey distributed to inmates, where both religiosity and negative behaviors were self-reported, where participation in the survey was voluntary, and where the response rate was 45% The Florida DOC s Kairos Horizons Study The Florida Department of Corrections, which ran a faith-based dorm, Kairos Horizons, at its Tomoka Correctional Institution, performed an unpublished study of the effectiveness of the program. 77 To be eligible for the dorm, an inmate had to have had no disciplinary reports in the previous six 68. O CONNOR ET AL., supra note 65, at Kent R. Kerley et al., Religiosity, Religious Participation, and Negative Prison Behaviors, 44 J. FOR SCI. STUDY RELIGION 443, 446 (2005). 70. Id. 71. See supra text accompanying note Kerley et al., supra note 69, at Id. 74. Id. at 450 tbl.2. The p-value of this difference was less than Id. 75. Id. The p-value of this difference was above Id. 76. Id. at BUREAU OF RESEARCH & DATA ANALYSIS, FLA. DEP T OF CORR., COMPARING TOMOKA CI S FAITH-BASED DORM (KAIROS HORIZONS) WITH NON-PARTICIPANTS (2000). Note that Kairos (this program and that described in text accompanying notes infra) and Kainos (the program described in text accompanying notes infra) are different programs. Jonathan Burnside, From Cursillo to Prison: The Story of Kairos, in BURNSIDE ET AL., supra note 58, at 34, 36. The correct name of the dorm is apparently Kairos Horizon, not Kairos Horizons. Jonathan Burnside, Navigating by the Heavens: Horizon Communities, in BURNSIDE ET AL., supra note 58, at 196, 196.

15 2011] Faith-Based Prisons 57 months. 78 The 59 inmates who spent the entire six-month program at the faith-based dorm were compared to 8 inmates who didn t complete the six months, 741 inmates at Tomoka who didn t participate at all, and 54,997 inmates at other Florida prisons. 79 (The comparison groups were also limited to inmates without disciplinary reports in the previous six months.) 80 Inmates who completed the six-month program had lower rates of disciplinary reports than non-participants or inmates at other Florida prisons; about 5% of completers received disciplinary reports, compared to 37.5% of non-completers, 17% of non-participants, and 12% of inmates at other prisons. 81 If to see the effect of participation rather than the effect of program completion we lump non-completers and completers together, the rate becomes about 9%, 82 which isn t significantly different from the rate among non-participants at Tomoka or at other prisons. 83 A similar faith-based program in England also reports greater disciplinary improvement among program participants Denny s Kairos Horizon Study Dan Denny analyzed in-prison misconduct and post-release recidivism rates for participants in a Kairos Horizon program at the Davis Correctional Facility, a private, medium-security prison in Oklahoma BUREAU OF RESEARCH & DATA ANALYSIS, supra note 77, at 2. However, four inmates were allowed into the program even though they had disciplinary reports in the previous six months. Id. 79. Id. at 2, Id. at See id. 82. Id. Of the 8 non-completers, 3 got disciplinary reports, while of the 59 completers, there were also 3 that got disciplinary reports. Id. Together, that makes 6 inmates receiving disciplinary reports out of 67 participants. Id. 83. The paper doesn t report statistical significance. But we can calculate it ourselves based on the table, id. at 9, using Pearson s chi-square test. Among the 67 participants, 6 inmates received disciplinary reports and 61 didn t. Among the 741 non-participants at Tomoka, 124 received disciplinary reports and 617 didn t (for a rate of 17%). Among the 54,997 non-participants at other prisons, 6,614 received disciplinary reports and 48,383 didn t (for a rate of 12%). The difference between the participants disciplinary report rate of 9% and the Tomoka non-participants rate of 17% has a p-value of 0.08, so it isn t significant at the 5% significance level. The difference between the participants rate of 9% and the other-prison non-participants rate of 12% has a p-value of 0.45, so it isn t significant at any reasonable significance level. 84. Id. at 13. There are also reports of an earlier study, conducted in 1995, evaluating the Kairos program at Union Correctional Institution in Florida. The study examined recidivism among 505 inmates who had attended Kairos over 10 years or who had attended 11 Kairos Weekends. The non-kairos control group had a 23.4% recidivism rate; the Kairos group had a 15.7% recidivism rate; and those who had participated in a Kairos follow-up program in addition to attending a Weekend had a recidivism rate of 10%. Burnside, From Cursillo to Prison, supra note 77, at 62 (citing Profile of Kairos, KAIROS NEWSL., (Kairos Prison Ministry, Winter Park, Fla.) Dec. 1, 1998); see also Kairos Fact Sheet, KAIROS PRISON MINISTRY INT L, 02 (last visited Oct. 6, 2011). However, I haven t been able to obtain this report (Kairos Prison Ministry International doesn t use it anymore, see from Ann M. Kreiler, Exec. Admin. Assistant, Kairos Prison Ministry Int l, to author (Jan. 26, 2011) (on file with author)), so it s unclear whether the results are statistically significant and whether they re tainted by self-selection. 85. Dan Denny, Individual and Organizational Impact of Kairos Horizon, a Faith-Based Adult Learning Program, in a Correctional Setting 41 (May 2006) (unpublished D.Ed. thesis, Oklahoma State University) (on file with the Oklahoma State University Library), available at

16 58 Alabama Law Review [Vol.63:1:43 Denny examined three cohorts of participants, from Year One (2002), Year Two (2003), and Year Three (2004). 86 The 36 Year One participants had 89% fewer misconduct reports after the program than before; the drop for the 51 Year Two participants was 80%; and the drop for the 51 Year Three participants was 84%. 87 The average drop was 86%. 88 Misconduct reports in the entire facility fell from 901 to 308 (a 66% drop) from Year One to Year Three, 89 which is presumably comparable to the 80% before-to-after drop for the Year Two participants. It s unclear from the paper how many inmates there were at the facility during this time, so it s unclear whether the drop in misconduct among program participants is significantly different from the total decrease facility-wide. When the paper was written, only seven participants had been released, the longest-released graduate had only been out for one year, and no graduate had been re-arrested. 90 So the author couldn t report true recidivism rates by Oklahoma standards, which require a three-year post-release history Education Studies Some education studies also use this approach, neither addressing selfselection nor controlling for observable variables. One example is Janet Beales and Maureen Wahl s assessment of the Partners Advancing Values in Evaluation (PAVE) program in Milwaukee, 92 a privately funded voucher system that functioned parallel to the publicly funded voucher system, the Milwaukee Parental Choice Program (MPCP). 93 Beales and Wahl found that 63.2% of PAVE students scored above the 50th percentile in reading (60.4% in math), which was much higher than the corresponding percentages for MPCP students, Milwaukee public school lowincome students, or all Milwaukee public school students. 94 (These percentages were all between 16% and 35%.) PAVE students were similarly above the three comparison groups in reading and math test score medians and means. 95 state.edu/etd/umi-okstate-1723.pdf. 86. Id. at Id. at 92 tbl Id. 89. Id. at 94 tbl Id. at Id. 92. Janet R. Beales & Maureen Wahl, Private Vouchers in Milwaukee: The PAVE Program, in PRIVATE VOUCHERS 41 (Terry M. Moe ed., 1995). Similar results are reported in JANET R. BEALES & MAUREEN WAHL, GIVEN THE CHOICE: A STUDY OF THE PAVE PROGRAM AND SCHOOL CHOICE IN MILWAUKEE (Reason Found. Pol y Study No. 183, Jan. 1995). 93. On the publicly funded voucher program in Milwaukee, see sources cited infra notes 151, , Beales & Wahl, supra note 92, at 61 tbl Id.

17 2011] Faith-Based Prisons 59 However, the PAVE group differed from the other groups in various ways. Most obviously, the PAVE group, like the MPCP group, was selfselected, since one had to apply for a voucher; the public school students weren t self-selected. 96 But the PAVE group and the MPCP group weren t comparable either: the PAVE scores were the test results of seventh-grade students, while the MPCP scores were test results from multiple grade levels, so the authors weren t even comparing the same test. 97 Finally, the authors couldn t control for income, parental education, or other variables. 98 B. Studies with Some Controls The studies in the previous section aren t credible because participants in religious programs are just so different from non-participants. One possible fix would be to control for observable differences between participants and non-participants. 99 This is what the studies reported in this section do: participants are matched with non-participants with observable characteristics that are as similar as possible. But these studies are still vulnerable. An unobserved variable motivation to change affects both whether the inmate participates and whether he reoffends. 100 Because motivation and success (avoiding rearrest) are positively correlated, any effect we find is probably biased upward (ignoring any other sources of bias in one direction or another). A true zero effect may look like a positive effect because we re measuring the effect of motivation. 101 In other words, if two prisoners are perfectly matched on the observables, but one of them chose to participate and the other didn t, these two prisoners aren t really well matched. Any study that finds better results among participants is thus still subject to self-selection bias La Vigne et al. s Florida Study Nancy La Vigne and her coauthors reported on six- and twelve-month recidivism rates of participants in two Florida faith- and character-based institutions (FCBI) one male (Lawtey) and one female (Hillsborough) Id. at 60. There may have been other self-selection biases: the PAVE test scores were voluntarily revealed by some of the participating parents, and perhaps parents would be more willing to reveal their children s test scores if they were high. 97. Id. at 61 tbl.10 nn.a-b. 98. Id. at Heckman & Robb, supra note E.g., Paul R. Rosenbaum & Donald B. Rubin, Reducing Bias in Observational Studies Using Subclassification on the Propensity Score, 79 J. AM. STAT. ASS N 516, 516 (1984); Heckman & Robb, supra note 50, at Cf. Caroline Minter Hoxby, Are Efficiency and Equity in School Finance Substitutes or Complements?, J. ECON. PERSP., Fall 1996, at 51, Moreover, the variables that are controlled for have no obvious connection to motivation, so their inclusion may not alleviate this particular selection problem. Mears et al., supra note 32, at NANCY G. LAVIGNE ET AL., EVALUATION OF FLORIDA S FAITH- AND CHARACTER-BASED

18 60 Alabama Law Review [Vol.63:1:43 Participants were matched with a control group based on sex, age, race, primary offense type, violent/non-violent offense, number of prior incarcerations, time incarcerated for current offense, time to expected release, and pre-study disciplinary report rate. 104 At first, male FCBI participants had lower recidivism rates than their control group none of the 189 male inmates from Lawtey were reincarcerated after six months, compared to four of the 189 male comparison inmates (2.1%). 105 There was no significant difference for females, 106 and twelve months out, there was no significant effect at all for either males or females. 107 There was also no significant difference between average time to reincarceration for the faith-based inmates and the comparison inmates, for either males or females. 108 The results here are thus extremely weak. 109 A later report by Diana Brazzell and Nancy La Vigne, using new data, continued to find no statistically significant difference... in the proportion of FCBI and non-fcbi inmates returned to prison within 12, 18, 24, and 26 months of release, for either males or females Rose s Kainos Community Study Gerry Rose evaluated the effect on reconviction of participation in the Kainos Community, a faith-based prison chiefly operating out of The Verne prison in England. 111 The 84 participants were compared against a sample of 13,832 prisoners; the comparison sample was composed of all adult sen- INSTITUTIONS: FINAL REPORT 1, (Urban Inst. Justice Pol y Ctr., Oct. 2007), /uploadedpdf/411561_fcbi_ evaluation.pdf Id Id. at 45 tbl.g. This was statistically significant at p Id Id. None of the 100 Hillsborough females were reincarcerated within 6 months, as opposed to one female in the comparison group. Id Id. at 45. Of the 56 male Lawtey inmates who were released at least a year before the study end date, one was reincarcerated, as opposed to two of the 82 in the comparison group. Id. Of the 54 female Hillsborough inmates who were released at least a year before the study end date, one was reincarcerated, as opposed to four of the 62 in the comparison group. Id Id. at Among reincarcerated males, mean time to reincarceration was 371 days for the Lawtey inmates and 262 days for the comparison group. Id. For females, the difference is 385 days versus 318 days. Id. Neither of these differences is significant at p<0.05. Id Id. at Diana Brazzell & Nancy La Vigne, Evaluating the Potential of Faith-Based Correctional Models: A Case Study of Florida s Faith-Based and Character-Based Institutions, presented at the Faith- Based and Community Initiatives Conference on Research, Outcomes, and Evaluation (June 2008), Gerry Rose, Kainos Community and Reconviction Rates, in JONATHAN BURNSIDE ET AL., KAINOS COMMUNITY IN PRISONS: REPORT OF AN EVALUATION 42 (presented to Res. Dev. & Stats. Directorate, Home Office, HM Prison Service England & Wales & Kainos Community, Dec. 2001), rep.pdf. Steve Aos and his coauthors considered this study to be one of the few that were of good enough quality to include in their review of evidence-based adult corrections programs. AOS ET AL., supra note 33, at 19. The program also had some participants at Highpoint North, Highpoint South, and Swaleside prisons, though there were apparently no Swaleside participants in Rose s empirical study. Rose, supra, at 34; Jonathan Burnside, Introduction, in BURNSIDE ET AL., supra, at 16. Substantially the same study was printed, with two years of follow-up data instead of one, as in Rose, supra note 63. There was no significant difference between the results of these two studies.

19 2011] Faith-Based Prisons 61 tenced prisoners released from prisons in England and Wales in 1996 and 1997 who were British nationals, had served sentences of six months to 15 years, had been released from particular categories of prisons, and satisfied a few additional restrictions. 112 In the Kainos sample, 22.6% of the participants were reconvicted within a year of release; among non-participants, the percentage was 25.9%. 113 This difference wasn t significant. 114 So far, this didn t control for any variables. But Rose then went further, comparing the actual reconviction rates of Kainos participants with their own predicted reconviction rates. The predicted rates were based on a statistical model that controlled for observable factors such as their sex, offense category, age at first conviction, age at sentence, months spent in prison after sentence, and number of custodial sentences before age Thus, rather than comparing participants and non-participants, he compared actual participants with hypothetical participants whose recidivism was predicted based on factors that didn t include their participation in a faith-based program. There, too, Rose found no significant effect: 25.0% of the Kainos sample was reconvicted, while the expected percentage would have been 26.0% or 24.2% (depending on which prediction model one used) Young et al. s Prison Ministry Study Mark Young and his coauthors investigated long-term recidivism among... federal inmates trained as volunteer prison ministers as part of Prison Fellowship Ministries Washington D.C. Discipleship Seminars. 117 Participants were sent to Washington for a two-week faith and leadership seminar, and their recidivism was compared to that of a control group. 118 The control group was selected to match the experimental group with respect to race, gender, age at release, and the salient factor score 119 (an estimate of a prisoner s likelihood of recidivism 120 ). Participants recidivism rate was 40%, while the control group s recidivism rate was 51%. 121 Participating women had a recidivism rate of 19%, compared to 47% for the control women, and participating men had a reci Rose, supra note 111, at Id. at 47 tbl Id. at 48 n.17. The p-value was close to Id Id. at 49, app Id. at 50. The same results held when the Kainos sample was divided by prison, and, within one of the prisons, by amount of exposure to the Kainos program. Id. The prediction model came in two forms a second version also took account of which prison the offender was released from. Id. at Mark C. Young et al., Long-Term Recidivism Among Federal Inmates Trained as Volunteer Prison Ministers, 22 J. OFFENDER REHAB. nos. 1 & 2, 1995, at 97. O Connor rates this study as having fair methodological quality (on a poor-fair-good-excellent scale). O Connor, supra note 33, at 23 tbl Young et al., supra note 117, at Id. at Peter B. Hoffman et al., Salient Factor Score and Releasee Behavior: Three Validation Samples, 2 LAW & HUM. BEHAV. 47, 49 tbl.1 (1978) Young et al., supra note 117, at 107. This difference was significant at p=0.04. Id. at 106.

20 62 Alabama Law Review [Vol.63:1:43 divism rate of 45%, compared to 52% for the control men. 122 When the groups were further broken down by gender and race, participants had lower recidivism rates for all subgroups except black men. 123 As in the theology study above, 124 these results are subject to both selfselection and selection by program administrators, in this case prison chaplains, who chose which inmates could participate O Connor et al. s Lieber Prison Study Tom O Connor and his coauthors reported on rates of in-prison infractions among participants in Prison Fellowship (PF) programming at Lieber Prison in South Carolina. 126 Their data set of 1,597 included both participants and non-participants; 302 inmates attended at least one out of 47 Prison Fellowship meetings. 127 Participants had lower infraction rates than non-participants: 9.9% of PF inmates had an infraction since attending at least one PF program compared to the 23.2% of Non PF inmates who had an infraction. 128 The more an inmate participated in PF programs, the lower his chance of having an infraction. 129 Controlling for prior violent convictions, age, marriage status, and days spent in the prison, whether an inmate participated in PF programs strongly predicted lower infraction rates. 130 Non PF inmates were still 2.5 times more likely than PF inmates to have an infraction. 131 The rate of participation in PF programs, controlling for the same variables, likewise strongly predicted lower infraction rates. 132 But controlling for the rate of participation isn t useful. Given a valid control group, the only valid comparison is between the control group and the entire treatment group. 133 If we compare the control group to isolated, self-selected subsets of the treatment group, like those who participated the most in PF programs, we are merely reintroducing another layer of self-selection bias. Even if high participation reduces infraction rates (which is doubtful, given that the high participants may already be better people), the relevant question from a policy perspective, that is, from the perspective of someone wondering 122. Id. at tbl Id See supra text accompanying notes Young et al., supra note 117, at THOMAS P. O CONNOR ET AL., THE IMPACT OF PRISON FELLOWSHIP ON INMATE INFRACTIONS AT LIEBER PRISON IN SOUTH CAROLINA (1997) Id. at 2, Id. at Id. at Id Id. The p-value was p< See id. at 9 10, tbl.1 (providing a Chi-square of and 5 degrees of freedom) Id. at 10. The p-value was p< See id. at 10, tbl.2. (providing a Chi-square of and 5 degrees of freedom) Camp et al., supra note 46, at 532.